4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
27 * Copyright (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
30 * Copyright (c) 2011, 2012, Intel Corporation.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
37 #define DEBUG_SUBSYSTEM S_LMV
38 #include <linux/slab.h>
39 #include <linux/module.h>
40 #include <linux/init.h>
41 #include <linux/pagemap.h>
43 #include <asm/div64.h>
44 #include <linux/seq_file.h>
45 #include <linux/namei.h>
46 #include <linux/uaccess.h>
48 #include "../include/lustre/lustre_idl.h"
49 #include "../include/obd_support.h"
50 #include "../include/lustre_lib.h"
51 #include "../include/lustre_net.h"
52 #include "../include/obd_class.h"
53 #include "../include/lprocfs_status.h"
54 #include "../include/lustre_lite.h"
55 #include "../include/lustre_fid.h"
56 #include "lmv_internal.h"
58 static void lmv_activate_target(struct lmv_obd *lmv,
59 struct lmv_tgt_desc *tgt,
62 if (tgt->ltd_active == activate)
65 tgt->ltd_active = activate;
66 lmv->desc.ld_active_tgt_count += (activate ? 1 : -1);
72 * -EINVAL : UUID can't be found in the LMV's target list
73 * -ENOTCONN: The UUID is found, but the target connection is bad (!)
74 * -EBADF : The UUID is found, but the OBD of the wrong type (!)
76 static int lmv_set_mdc_active(struct lmv_obd *lmv, struct obd_uuid *uuid,
79 struct lmv_tgt_desc *uninitialized_var(tgt);
80 struct obd_device *obd;
84 CDEBUG(D_INFO, "Searching in lmv %p for uuid %s (activate=%d)\n",
85 lmv, uuid->uuid, activate);
87 spin_lock(&lmv->lmv_lock);
88 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
90 if (tgt == NULL || tgt->ltd_exp == NULL)
93 CDEBUG(D_INFO, "Target idx %d is %s conn %#llx\n", i,
94 tgt->ltd_uuid.uuid, tgt->ltd_exp->exp_handle.h_cookie);
96 if (obd_uuid_equals(uuid, &tgt->ltd_uuid))
100 if (i == lmv->desc.ld_tgt_count) {
105 obd = class_exp2obd(tgt->ltd_exp);
111 CDEBUG(D_INFO, "Found OBD %s=%s device %d (%p) type %s at LMV idx %d\n",
112 obd->obd_name, obd->obd_uuid.uuid, obd->obd_minor, obd,
113 obd->obd_type->typ_name, i);
114 LASSERT(strcmp(obd->obd_type->typ_name, LUSTRE_MDC_NAME) == 0);
116 if (tgt->ltd_active == activate) {
117 CDEBUG(D_INFO, "OBD %p already %sactive!\n", obd,
118 activate ? "" : "in");
122 CDEBUG(D_INFO, "Marking OBD %p %sactive\n", obd,
123 activate ? "" : "in");
124 lmv_activate_target(lmv, tgt, activate);
127 spin_unlock(&lmv->lmv_lock);
131 static struct obd_uuid *lmv_get_uuid(struct obd_export *exp)
133 struct lmv_obd *lmv = &exp->exp_obd->u.lmv;
135 return obd_get_uuid(lmv->tgts[0]->ltd_exp);
138 static int lmv_notify(struct obd_device *obd, struct obd_device *watched,
139 enum obd_notify_event ev, void *data)
141 struct obd_connect_data *conn_data;
142 struct lmv_obd *lmv = &obd->u.lmv;
143 struct obd_uuid *uuid;
146 if (strcmp(watched->obd_type->typ_name, LUSTRE_MDC_NAME)) {
147 CERROR("unexpected notification of %s %s!\n",
148 watched->obd_type->typ_name,
153 uuid = &watched->u.cli.cl_target_uuid;
154 if (ev == OBD_NOTIFY_ACTIVE || ev == OBD_NOTIFY_INACTIVE) {
156 * Set MDC as active before notifying the observer, so the
157 * observer can use the MDC normally.
159 rc = lmv_set_mdc_active(lmv, uuid,
160 ev == OBD_NOTIFY_ACTIVE);
162 CERROR("%sactivation of %s failed: %d\n",
163 ev == OBD_NOTIFY_ACTIVE ? "" : "de",
167 } else if (ev == OBD_NOTIFY_OCD) {
168 conn_data = &watched->u.cli.cl_import->imp_connect_data;
170 * XXX: Make sure that ocd_connect_flags from all targets are
171 * the same. Otherwise one of MDTs runs wrong version or
172 * something like this. --umka
174 obd->obd_self_export->exp_connect_data = *conn_data;
177 else if (ev == OBD_NOTIFY_DISCON) {
179 * For disconnect event, flush fld cache for failout MDS case.
181 fld_client_flush(&lmv->lmv_fld);
185 * Pass the notification up the chain.
187 if (obd->obd_observer)
188 rc = obd_notify(obd->obd_observer, watched, ev, data);
194 * This is fake connect function. Its purpose is to initialize lmv and say
195 * caller that everything is okay. Real connection will be performed later.
197 static int lmv_connect(const struct lu_env *env,
198 struct obd_export **exp, struct obd_device *obd,
199 struct obd_uuid *cluuid, struct obd_connect_data *data,
202 struct lmv_obd *lmv = &obd->u.lmv;
203 struct lustre_handle conn = { 0 };
207 * We don't want to actually do the underlying connections more than
208 * once, so keep track.
211 if (lmv->refcount > 1) {
216 rc = class_connect(&conn, obd, cluuid);
218 CERROR("class_connection() returned %d\n", rc);
222 *exp = class_conn2export(&conn);
223 class_export_get(*exp);
227 lmv->cluuid = *cluuid;
230 lmv->conn_data = *data;
232 lmv->lmv_tgts_kobj = kobject_create_and_add("target_obds",
235 * All real clients should perform actual connection right away, because
236 * it is possible, that LMV will not have opportunity to connect targets
237 * and MDC stuff will be called directly, for instance while reading
238 * ../mdc/../kbytesfree procfs file, etc.
240 if (data->ocd_connect_flags & OBD_CONNECT_REAL)
241 rc = lmv_check_connect(obd);
243 if (rc && lmv->lmv_tgts_kobj)
244 kobject_put(lmv->lmv_tgts_kobj);
249 static void lmv_set_timeouts(struct obd_device *obd)
251 struct lmv_tgt_desc *tgt;
256 if (lmv->server_timeout == 0)
259 if (lmv->connected == 0)
262 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
264 if (tgt == NULL || tgt->ltd_exp == NULL || tgt->ltd_active == 0)
267 obd_set_info_async(NULL, tgt->ltd_exp, sizeof(KEY_INTERMDS),
268 KEY_INTERMDS, 0, NULL, NULL);
272 static int lmv_init_ea_size(struct obd_export *exp, int easize,
273 int def_easize, int cookiesize, int def_cookiesize)
275 struct obd_device *obd = exp->exp_obd;
276 struct lmv_obd *lmv = &obd->u.lmv;
281 if (lmv->max_easize < easize) {
282 lmv->max_easize = easize;
285 if (lmv->max_def_easize < def_easize) {
286 lmv->max_def_easize = def_easize;
289 if (lmv->max_cookiesize < cookiesize) {
290 lmv->max_cookiesize = cookiesize;
293 if (lmv->max_def_cookiesize < def_cookiesize) {
294 lmv->max_def_cookiesize = def_cookiesize;
300 if (lmv->connected == 0)
303 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
304 if (lmv->tgts[i] == NULL ||
305 lmv->tgts[i]->ltd_exp == NULL ||
306 lmv->tgts[i]->ltd_active == 0) {
307 CWARN("%s: NULL export for %d\n", obd->obd_name, i);
311 rc = md_init_ea_size(lmv->tgts[i]->ltd_exp, easize, def_easize,
312 cookiesize, def_cookiesize);
314 CERROR("%s: obd_init_ea_size() failed on MDT target %d: rc = %d.\n",
315 obd->obd_name, i, rc);
322 #define MAX_STRING_SIZE 128
324 static int lmv_connect_mdc(struct obd_device *obd, struct lmv_tgt_desc *tgt)
326 struct lmv_obd *lmv = &obd->u.lmv;
327 struct obd_uuid *cluuid = &lmv->cluuid;
328 struct obd_uuid lmv_mdc_uuid = { "LMV_MDC_UUID" };
329 struct obd_device *mdc_obd;
330 struct obd_export *mdc_exp;
331 struct lu_fld_target target;
334 mdc_obd = class_find_client_obd(&tgt->ltd_uuid, LUSTRE_MDC_NAME,
337 CERROR("target %s not attached\n", tgt->ltd_uuid.uuid);
341 CDEBUG(D_CONFIG, "connect to %s(%s) - %s, %s FOR %s\n",
342 mdc_obd->obd_name, mdc_obd->obd_uuid.uuid,
343 tgt->ltd_uuid.uuid, obd->obd_uuid.uuid,
346 if (!mdc_obd->obd_set_up) {
347 CERROR("target %s is not set up\n", tgt->ltd_uuid.uuid);
351 rc = obd_connect(NULL, &mdc_exp, mdc_obd, &lmv_mdc_uuid,
352 &lmv->conn_data, NULL);
354 CERROR("target %s connect error %d\n", tgt->ltd_uuid.uuid, rc);
359 * Init fid sequence client for this mdc and add new fld target.
361 rc = obd_fid_init(mdc_obd, mdc_exp, LUSTRE_SEQ_METADATA);
365 target.ft_srv = NULL;
366 target.ft_exp = mdc_exp;
367 target.ft_idx = tgt->ltd_idx;
369 fld_client_add_target(&lmv->lmv_fld, &target);
371 rc = obd_register_observer(mdc_obd, obd);
373 obd_disconnect(mdc_exp);
374 CERROR("target %s register_observer error %d\n",
375 tgt->ltd_uuid.uuid, rc);
379 if (obd->obd_observer) {
381 * Tell the observer about the new target.
383 rc = obd_notify(obd->obd_observer, mdc_exp->exp_obd,
385 (void *)(tgt - lmv->tgts[0]));
387 obd_disconnect(mdc_exp);
393 tgt->ltd_exp = mdc_exp;
394 lmv->desc.ld_active_tgt_count++;
396 md_init_ea_size(tgt->ltd_exp, lmv->max_easize, lmv->max_def_easize,
397 lmv->max_cookiesize, lmv->max_def_cookiesize);
399 CDEBUG(D_CONFIG, "Connected to %s(%s) successfully (%d)\n",
400 mdc_obd->obd_name, mdc_obd->obd_uuid.uuid,
401 atomic_read(&obd->obd_refcount));
403 if (lmv->lmv_tgts_kobj)
404 /* Even if we failed to create the link, that's fine */
405 rc = sysfs_create_link(lmv->lmv_tgts_kobj, &mdc_obd->obd_kobj,
410 static void lmv_del_target(struct lmv_obd *lmv, int index)
412 if (lmv->tgts[index] == NULL)
415 kfree(lmv->tgts[index]);
416 lmv->tgts[index] = NULL;
420 static int lmv_add_target(struct obd_device *obd, struct obd_uuid *uuidp,
421 __u32 index, int gen)
423 struct lmv_obd *lmv = &obd->u.lmv;
424 struct lmv_tgt_desc *tgt;
427 CDEBUG(D_CONFIG, "Target uuid: %s. index %d\n", uuidp->uuid, index);
431 if (lmv->desc.ld_tgt_count == 0) {
432 struct obd_device *mdc_obd;
434 mdc_obd = class_find_client_obd(uuidp, LUSTRE_MDC_NAME,
437 lmv_init_unlock(lmv);
438 CERROR("%s: Target %s not attached: rc = %d\n",
439 obd->obd_name, uuidp->uuid, -EINVAL);
444 if ((index < lmv->tgts_size) && (lmv->tgts[index] != NULL)) {
445 tgt = lmv->tgts[index];
446 CERROR("%s: UUID %s already assigned at LOV target index %d: rc = %d\n",
448 obd_uuid2str(&tgt->ltd_uuid), index, -EEXIST);
449 lmv_init_unlock(lmv);
453 if (index >= lmv->tgts_size) {
454 /* We need to reallocate the lmv target array. */
455 struct lmv_tgt_desc **newtgts, **old = NULL;
459 while (newsize < index + 1)
461 newtgts = kcalloc(newsize, sizeof(*newtgts), GFP_NOFS);
462 if (newtgts == NULL) {
463 lmv_init_unlock(lmv);
467 if (lmv->tgts_size) {
468 memcpy(newtgts, lmv->tgts,
469 sizeof(*newtgts) * lmv->tgts_size);
471 oldsize = lmv->tgts_size;
475 lmv->tgts_size = newsize;
479 CDEBUG(D_CONFIG, "tgts: %p size: %d\n", lmv->tgts,
483 tgt = kzalloc(sizeof(*tgt), GFP_NOFS);
485 lmv_init_unlock(lmv);
489 mutex_init(&tgt->ltd_fid_mutex);
490 tgt->ltd_idx = index;
491 tgt->ltd_uuid = *uuidp;
493 lmv->tgts[index] = tgt;
494 if (index >= lmv->desc.ld_tgt_count)
495 lmv->desc.ld_tgt_count = index + 1;
497 if (lmv->connected) {
498 rc = lmv_connect_mdc(obd, tgt);
500 spin_lock(&lmv->lmv_lock);
501 lmv->desc.ld_tgt_count--;
502 memset(tgt, 0, sizeof(*tgt));
503 spin_unlock(&lmv->lmv_lock);
505 int easize = sizeof(struct lmv_stripe_md) +
506 lmv->desc.ld_tgt_count * sizeof(struct lu_fid);
507 lmv_init_ea_size(obd->obd_self_export, easize, 0, 0, 0);
511 lmv_init_unlock(lmv);
515 int lmv_check_connect(struct obd_device *obd)
517 struct lmv_obd *lmv = &obd->u.lmv;
518 struct lmv_tgt_desc *tgt;
527 if (lmv->connected) {
528 lmv_init_unlock(lmv);
532 if (lmv->desc.ld_tgt_count == 0) {
533 lmv_init_unlock(lmv);
534 CERROR("%s: no targets configured.\n", obd->obd_name);
538 CDEBUG(D_CONFIG, "Time to connect %s to %s\n",
539 lmv->cluuid.uuid, obd->obd_name);
541 LASSERT(lmv->tgts != NULL);
543 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
547 rc = lmv_connect_mdc(obd, tgt);
552 lmv_set_timeouts(obd);
553 class_export_put(lmv->exp);
555 easize = lmv_get_easize(lmv);
556 lmv_init_ea_size(obd->obd_self_export, easize, 0, 0, 0);
557 lmv_init_unlock(lmv);
569 --lmv->desc.ld_active_tgt_count;
570 rc2 = obd_disconnect(tgt->ltd_exp);
572 CERROR("LMV target %s disconnect on MDC idx %d: error %d\n",
573 tgt->ltd_uuid.uuid, i, rc2);
577 class_disconnect(lmv->exp);
578 lmv_init_unlock(lmv);
582 static int lmv_disconnect_mdc(struct obd_device *obd, struct lmv_tgt_desc *tgt)
584 struct lmv_obd *lmv = &obd->u.lmv;
585 struct obd_device *mdc_obd;
588 LASSERT(tgt != NULL);
589 LASSERT(obd != NULL);
591 mdc_obd = class_exp2obd(tgt->ltd_exp);
594 mdc_obd->obd_force = obd->obd_force;
595 mdc_obd->obd_fail = obd->obd_fail;
596 mdc_obd->obd_no_recov = obd->obd_no_recov;
598 if (lmv->lmv_tgts_kobj)
599 sysfs_remove_link(lmv->lmv_tgts_kobj,
603 rc = obd_fid_fini(tgt->ltd_exp->exp_obd);
605 CERROR("Can't finalize fids factory\n");
607 CDEBUG(D_INFO, "Disconnected from %s(%s) successfully\n",
608 tgt->ltd_exp->exp_obd->obd_name,
609 tgt->ltd_exp->exp_obd->obd_uuid.uuid);
611 obd_register_observer(tgt->ltd_exp->exp_obd, NULL);
612 rc = obd_disconnect(tgt->ltd_exp);
614 if (tgt->ltd_active) {
615 CERROR("Target %s disconnect error %d\n",
616 tgt->ltd_uuid.uuid, rc);
620 lmv_activate_target(lmv, tgt, 0);
625 static int lmv_disconnect(struct obd_export *exp)
627 struct obd_device *obd = class_exp2obd(exp);
628 struct lmv_obd *lmv = &obd->u.lmv;
636 * Only disconnect the underlying layers on the final disconnect.
639 if (lmv->refcount != 0)
642 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
643 if (lmv->tgts[i] == NULL || lmv->tgts[i]->ltd_exp == NULL)
646 lmv_disconnect_mdc(obd, lmv->tgts[i]);
649 if (lmv->lmv_tgts_kobj)
650 kobject_put(lmv->lmv_tgts_kobj);
654 * This is the case when no real connection is established by
655 * lmv_check_connect().
658 class_export_put(exp);
659 rc = class_disconnect(exp);
660 if (lmv->refcount == 0)
665 static int lmv_fid2path(struct obd_export *exp, int len, void *karg, void *uarg)
667 struct obd_device *obddev = class_exp2obd(exp);
668 struct lmv_obd *lmv = &obddev->u.lmv;
669 struct getinfo_fid2path *gf;
670 struct lmv_tgt_desc *tgt;
671 struct getinfo_fid2path *remote_gf = NULL;
672 int remote_gf_size = 0;
675 gf = (struct getinfo_fid2path *)karg;
676 tgt = lmv_find_target(lmv, &gf->gf_fid);
681 rc = obd_iocontrol(OBD_IOC_FID2PATH, tgt->ltd_exp, len, gf, uarg);
682 if (rc != 0 && rc != -EREMOTE)
685 /* If remote_gf != NULL, it means just building the
686 * path on the remote MDT, copy this path segment to gf */
687 if (remote_gf != NULL) {
688 struct getinfo_fid2path *ori_gf;
691 ori_gf = (struct getinfo_fid2path *)karg;
692 if (strlen(ori_gf->gf_path) +
693 strlen(gf->gf_path) > ori_gf->gf_pathlen) {
698 ptr = ori_gf->gf_path;
700 memmove(ptr + strlen(gf->gf_path) + 1, ptr,
701 strlen(ori_gf->gf_path));
703 strcpy(ptr, gf->gf_path);
704 ptr += strlen(gf->gf_path);
708 CDEBUG(D_INFO, "%s: get path %s "DFID" rec: %llu ln: %u\n",
709 tgt->ltd_exp->exp_obd->obd_name,
710 gf->gf_path, PFID(&gf->gf_fid), gf->gf_recno,
716 /* sigh, has to go to another MDT to do path building further */
717 if (remote_gf == NULL) {
718 remote_gf_size = sizeof(*remote_gf) + PATH_MAX;
719 remote_gf = kzalloc(remote_gf_size, GFP_NOFS);
724 remote_gf->gf_pathlen = PATH_MAX;
727 if (!fid_is_sane(&gf->gf_fid)) {
728 CERROR("%s: invalid FID "DFID": rc = %d\n",
729 tgt->ltd_exp->exp_obd->obd_name,
730 PFID(&gf->gf_fid), -EINVAL);
735 tgt = lmv_find_target(lmv, &gf->gf_fid);
741 remote_gf->gf_fid = gf->gf_fid;
742 remote_gf->gf_recno = -1;
743 remote_gf->gf_linkno = -1;
744 memset(remote_gf->gf_path, 0, remote_gf->gf_pathlen);
746 goto repeat_fid2path;
753 static int lmv_hsm_req_count(struct lmv_obd *lmv,
754 const struct hsm_user_request *hur,
755 const struct lmv_tgt_desc *tgt_mds)
758 struct lmv_tgt_desc *curr_tgt;
760 /* count how many requests must be sent to the given target */
761 for (i = 0; i < hur->hur_request.hr_itemcount; i++) {
762 curr_tgt = lmv_find_target(lmv, &hur->hur_user_item[i].hui_fid);
763 if (obd_uuid_equals(&curr_tgt->ltd_uuid, &tgt_mds->ltd_uuid))
769 static void lmv_hsm_req_build(struct lmv_obd *lmv,
770 struct hsm_user_request *hur_in,
771 const struct lmv_tgt_desc *tgt_mds,
772 struct hsm_user_request *hur_out)
775 struct lmv_tgt_desc *curr_tgt;
777 /* build the hsm_user_request for the given target */
778 hur_out->hur_request = hur_in->hur_request;
780 for (i = 0; i < hur_in->hur_request.hr_itemcount; i++) {
781 curr_tgt = lmv_find_target(lmv,
782 &hur_in->hur_user_item[i].hui_fid);
783 if (obd_uuid_equals(&curr_tgt->ltd_uuid, &tgt_mds->ltd_uuid)) {
784 hur_out->hur_user_item[nr_out] =
785 hur_in->hur_user_item[i];
789 hur_out->hur_request.hr_itemcount = nr_out;
790 memcpy(hur_data(hur_out), hur_data(hur_in),
791 hur_in->hur_request.hr_data_len);
794 static int lmv_hsm_ct_unregister(struct lmv_obd *lmv, unsigned int cmd, int len,
795 struct lustre_kernelcomm *lk, void *uarg)
799 /* unregister request (call from llapi_hsm_copytool_fini) */
800 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
801 /* best effort: try to clean as much as possible
802 * (continue on error) */
803 obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp, len, lk, uarg);
806 /* Whatever the result, remove copytool from kuc groups.
807 * Unreached coordinators will get EPIPE on next requests
808 * and will unregister automatically.
810 rc = libcfs_kkuc_group_rem(lk->lk_uid, lk->lk_group);
814 static int lmv_hsm_ct_register(struct lmv_obd *lmv, unsigned int cmd, int len,
815 struct lustre_kernelcomm *lk, void *uarg)
820 bool any_set = false;
822 /* All or nothing: try to register to all MDS.
823 * In case of failure, unregister from previous MDS,
824 * except if it because of inactive target. */
825 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
826 err = obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp,
829 if (lmv->tgts[i]->ltd_active) {
830 /* permanent error */
831 CERROR("error: iocontrol MDC %s on MDTidx %d cmd %x: err = %d\n",
832 lmv->tgts[i]->ltd_uuid.uuid,
835 lk->lk_flags |= LK_FLG_STOP;
836 /* unregister from previous MDS */
837 for (j = 0; j < i; j++)
839 lmv->tgts[j]->ltd_exp,
843 /* else: transient error.
844 * kuc will register to the missing MDT
852 /* no registration done: return error */
855 /* at least one registration done, with no failure */
856 filp = fget(lk->lk_wfd);
860 rc = libcfs_kkuc_group_add(filp, lk->lk_uid, lk->lk_group, lk->lk_data);
861 if (rc != 0 && filp != NULL)
866 static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,
867 int len, void *karg, void *uarg)
869 struct obd_device *obddev = class_exp2obd(exp);
870 struct lmv_obd *lmv = &obddev->u.lmv;
874 int count = lmv->desc.ld_tgt_count;
880 case IOC_OBD_STATFS: {
881 struct obd_ioctl_data *data = karg;
882 struct obd_device *mdc_obd;
883 struct obd_statfs stat_buf = {0};
886 memcpy(&index, data->ioc_inlbuf2, sizeof(__u32));
890 if (lmv->tgts[index] == NULL ||
891 lmv->tgts[index]->ltd_active == 0)
894 mdc_obd = class_exp2obd(lmv->tgts[index]->ltd_exp);
899 if (copy_to_user(data->ioc_pbuf2, obd2cli_tgt(mdc_obd),
900 min((int) data->ioc_plen2,
901 (int) sizeof(struct obd_uuid))))
904 rc = obd_statfs(NULL, lmv->tgts[index]->ltd_exp, &stat_buf,
905 cfs_time_shift_64(-OBD_STATFS_CACHE_SECONDS),
909 if (copy_to_user(data->ioc_pbuf1, &stat_buf,
910 min((int) data->ioc_plen1,
911 (int) sizeof(stat_buf))))
915 case OBD_IOC_QUOTACTL: {
916 struct if_quotactl *qctl = karg;
917 struct lmv_tgt_desc *tgt = NULL;
918 struct obd_quotactl *oqctl;
920 if (qctl->qc_valid == QC_MDTIDX) {
921 if (qctl->qc_idx < 0 || count <= qctl->qc_idx)
924 tgt = lmv->tgts[qctl->qc_idx];
925 if (tgt == NULL || tgt->ltd_exp == NULL)
927 } else if (qctl->qc_valid == QC_UUID) {
928 for (i = 0; i < count; i++) {
932 if (!obd_uuid_equals(&tgt->ltd_uuid,
936 if (tgt->ltd_exp == NULL)
948 LASSERT(tgt && tgt->ltd_exp);
949 oqctl = kzalloc(sizeof(*oqctl), GFP_NOFS);
953 QCTL_COPY(oqctl, qctl);
954 rc = obd_quotactl(tgt->ltd_exp, oqctl);
956 QCTL_COPY(qctl, oqctl);
957 qctl->qc_valid = QC_MDTIDX;
958 qctl->obd_uuid = tgt->ltd_uuid;
963 case OBD_IOC_CHANGELOG_SEND:
964 case OBD_IOC_CHANGELOG_CLEAR: {
965 struct ioc_changelog *icc = karg;
967 if (icc->icc_mdtindex >= count)
970 if (lmv->tgts[icc->icc_mdtindex] == NULL ||
971 lmv->tgts[icc->icc_mdtindex]->ltd_exp == NULL ||
972 lmv->tgts[icc->icc_mdtindex]->ltd_active == 0)
974 rc = obd_iocontrol(cmd, lmv->tgts[icc->icc_mdtindex]->ltd_exp,
975 sizeof(*icc), icc, NULL);
978 case LL_IOC_GET_CONNECT_FLAGS: {
979 if (lmv->tgts[0] == NULL)
981 rc = obd_iocontrol(cmd, lmv->tgts[0]->ltd_exp, len, karg, uarg);
984 case OBD_IOC_FID2PATH: {
985 rc = lmv_fid2path(exp, len, karg, uarg);
988 case LL_IOC_HSM_STATE_GET:
989 case LL_IOC_HSM_STATE_SET:
990 case LL_IOC_HSM_ACTION: {
991 struct md_op_data *op_data = karg;
992 struct lmv_tgt_desc *tgt;
994 tgt = lmv_find_target(lmv, &op_data->op_fid1);
998 if (tgt->ltd_exp == NULL)
1001 rc = obd_iocontrol(cmd, tgt->ltd_exp, len, karg, uarg);
1004 case LL_IOC_HSM_PROGRESS: {
1005 const struct hsm_progress_kernel *hpk = karg;
1006 struct lmv_tgt_desc *tgt;
1008 tgt = lmv_find_target(lmv, &hpk->hpk_fid);
1010 return PTR_ERR(tgt);
1011 rc = obd_iocontrol(cmd, tgt->ltd_exp, len, karg, uarg);
1014 case LL_IOC_HSM_REQUEST: {
1015 struct hsm_user_request *hur = karg;
1016 struct lmv_tgt_desc *tgt;
1017 unsigned int reqcount = hur->hur_request.hr_itemcount;
1022 /* if the request is about a single fid
1023 * or if there is a single MDS, no need to split
1025 if (reqcount == 1 || count == 1) {
1026 tgt = lmv_find_target(lmv,
1027 &hur->hur_user_item[0].hui_fid);
1029 return PTR_ERR(tgt);
1030 rc = obd_iocontrol(cmd, tgt->ltd_exp, len, karg, uarg);
1032 /* split fid list to their respective MDS */
1033 for (i = 0; i < count; i++) {
1034 unsigned int nr, reqlen;
1036 struct hsm_user_request *req;
1038 nr = lmv_hsm_req_count(lmv, hur, lmv->tgts[i]);
1039 if (nr == 0) /* nothing for this MDS */
1042 /* build a request with fids for this MDS */
1043 reqlen = offsetof(typeof(*hur),
1045 + hur->hur_request.hr_data_len;
1046 req = libcfs_kvzalloc(reqlen, GFP_NOFS);
1050 lmv_hsm_req_build(lmv, hur, lmv->tgts[i], req);
1052 rc1 = obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp,
1054 if (rc1 != 0 && rc == 0)
1061 case LL_IOC_LOV_SWAP_LAYOUTS: {
1062 struct md_op_data *op_data = karg;
1063 struct lmv_tgt_desc *tgt1, *tgt2;
1065 tgt1 = lmv_find_target(lmv, &op_data->op_fid1);
1067 return PTR_ERR(tgt1);
1069 tgt2 = lmv_find_target(lmv, &op_data->op_fid2);
1071 return PTR_ERR(tgt2);
1073 if ((tgt1->ltd_exp == NULL) || (tgt2->ltd_exp == NULL))
1076 /* only files on same MDT can have their layouts swapped */
1077 if (tgt1->ltd_idx != tgt2->ltd_idx)
1080 rc = obd_iocontrol(cmd, tgt1->ltd_exp, len, karg, uarg);
1083 case LL_IOC_HSM_CT_START: {
1084 struct lustre_kernelcomm *lk = karg;
1086 if (lk->lk_flags & LK_FLG_STOP)
1087 rc = lmv_hsm_ct_unregister(lmv, cmd, len, lk, uarg);
1089 rc = lmv_hsm_ct_register(lmv, cmd, len, lk, uarg);
1093 for (i = 0; i < count; i++) {
1094 struct obd_device *mdc_obd;
1097 if (lmv->tgts[i] == NULL ||
1098 lmv->tgts[i]->ltd_exp == NULL)
1100 /* ll_umount_begin() sets force flag but for lmv, not
1101 * mdc. Let's pass it through */
1102 mdc_obd = class_exp2obd(lmv->tgts[i]->ltd_exp);
1103 mdc_obd->obd_force = obddev->obd_force;
1104 err = obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp, len,
1106 if (err == -ENODATA && cmd == OBD_IOC_POLL_QUOTACHECK) {
1109 if (lmv->tgts[i]->ltd_active) {
1110 CERROR("error: iocontrol MDC %s on MDTidx %d cmd %x: err = %d\n",
1111 lmv->tgts[i]->ltd_uuid.uuid,
1126 static int lmv_all_chars_policy(int count, const char *name,
1137 static int lmv_nid_policy(struct lmv_obd *lmv)
1139 struct obd_import *imp;
1143 * XXX: To get nid we assume that underlying obd device is mdc.
1145 imp = class_exp2cliimp(lmv->tgts[0].ltd_exp);
1146 id = imp->imp_connection->c_self ^ (imp->imp_connection->c_self >> 32);
1147 return id % lmv->desc.ld_tgt_count;
1150 static int lmv_choose_mds(struct lmv_obd *lmv, struct md_op_data *op_data,
1151 enum placement_policy placement)
1153 switch (placement) {
1154 case PLACEMENT_CHAR_POLICY:
1155 return lmv_all_chars_policy(lmv->desc.ld_tgt_count,
1157 op_data->op_namelen);
1158 case PLACEMENT_NID_POLICY:
1159 return lmv_nid_policy(lmv);
1165 CERROR("Unsupported placement policy %x\n", placement);
1171 * This is _inode_ placement policy function (not name).
1173 static int lmv_placement_policy(struct obd_device *obd,
1174 struct md_op_data *op_data, u32 *mds)
1176 struct lmv_obd *lmv = &obd->u.lmv;
1178 LASSERT(mds != NULL);
1180 if (lmv->desc.ld_tgt_count == 1) {
1186 * If stripe_offset is provided during setdirstripe
1187 * (setdirstripe -i xx), xx MDS will be chosen.
1189 if (op_data->op_cli_flags & CLI_SET_MEA) {
1190 struct lmv_user_md *lum;
1192 lum = (struct lmv_user_md *)op_data->op_data;
1193 if (lum->lum_type == LMV_STRIPE_TYPE &&
1194 lum->lum_stripe_offset != -1) {
1195 if (lum->lum_stripe_offset >= lmv->desc.ld_tgt_count) {
1196 CERROR("%s: Stripe_offset %d > MDT count %d: rc = %d\n",
1198 lum->lum_stripe_offset,
1199 lmv->desc.ld_tgt_count, -ERANGE);
1202 *mds = lum->lum_stripe_offset;
1207 /* Allocate new fid on target according to operation type and parent
1209 *mds = op_data->op_mds;
1213 int __lmv_fid_alloc(struct lmv_obd *lmv, struct lu_fid *fid, u32 mds)
1215 struct lmv_tgt_desc *tgt;
1218 tgt = lmv_get_target(lmv, mds);
1220 return PTR_ERR(tgt);
1223 * New seq alloc and FLD setup should be atomic. Otherwise we may find
1224 * on server that seq in new allocated fid is not yet known.
1226 mutex_lock(&tgt->ltd_fid_mutex);
1228 if (tgt->ltd_active == 0 || tgt->ltd_exp == NULL) {
1234 * Asking underlaying tgt layer to allocate new fid.
1236 rc = obd_fid_alloc(tgt->ltd_exp, fid, NULL);
1238 LASSERT(fid_is_sane(fid));
1243 mutex_unlock(&tgt->ltd_fid_mutex);
1247 int lmv_fid_alloc(struct obd_export *exp, struct lu_fid *fid,
1248 struct md_op_data *op_data)
1250 struct obd_device *obd = class_exp2obd(exp);
1251 struct lmv_obd *lmv = &obd->u.lmv;
1255 LASSERT(op_data != NULL);
1256 LASSERT(fid != NULL);
1258 rc = lmv_placement_policy(obd, op_data, &mds);
1260 CERROR("Can't get target for allocating fid, rc %d\n",
1265 rc = __lmv_fid_alloc(lmv, fid, mds);
1267 CERROR("Can't alloc new fid, rc %d\n", rc);
1274 static int lmv_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
1276 struct lmv_obd *lmv = &obd->u.lmv;
1277 struct lprocfs_static_vars lvars = { NULL };
1278 struct lmv_desc *desc;
1281 if (LUSTRE_CFG_BUFLEN(lcfg, 1) < 1) {
1282 CERROR("LMV setup requires a descriptor\n");
1286 desc = (struct lmv_desc *)lustre_cfg_buf(lcfg, 1);
1287 if (sizeof(*desc) > LUSTRE_CFG_BUFLEN(lcfg, 1)) {
1288 CERROR("Lmv descriptor size wrong: %d > %d\n",
1289 (int)sizeof(*desc), LUSTRE_CFG_BUFLEN(lcfg, 1));
1293 lmv->tgts = kcalloc(32, sizeof(*lmv->tgts), GFP_NOFS);
1294 if (lmv->tgts == NULL)
1296 lmv->tgts_size = 32;
1298 obd_str2uuid(&lmv->desc.ld_uuid, desc->ld_uuid.uuid);
1299 lmv->desc.ld_tgt_count = 0;
1300 lmv->desc.ld_active_tgt_count = 0;
1301 lmv->max_cookiesize = 0;
1302 lmv->max_def_easize = 0;
1303 lmv->max_easize = 0;
1304 lmv->lmv_placement = PLACEMENT_CHAR_POLICY;
1306 spin_lock_init(&lmv->lmv_lock);
1307 mutex_init(&lmv->init_mutex);
1309 lprocfs_lmv_init_vars(&lvars);
1311 lprocfs_obd_setup(obd, lvars.obd_vars, lvars.sysfs_vars);
1312 rc = ldebugfs_seq_create(obd->obd_debugfs_entry, "target_obd",
1313 0444, &lmv_proc_target_fops, obd);
1315 CWARN("%s: error adding LMV target_obd file: rc = %d\n",
1317 rc = fld_client_init(&lmv->lmv_fld, obd->obd_name,
1318 LUSTRE_CLI_FLD_HASH_DHT);
1320 CERROR("Can't init FLD, err %d\n", rc);
1330 static int lmv_cleanup(struct obd_device *obd)
1332 struct lmv_obd *lmv = &obd->u.lmv;
1334 fld_client_fini(&lmv->lmv_fld);
1335 if (lmv->tgts != NULL) {
1338 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
1339 if (lmv->tgts[i] == NULL)
1341 lmv_del_target(lmv, i);
1349 static int lmv_process_config(struct obd_device *obd, u32 len, void *buf)
1351 struct lustre_cfg *lcfg = buf;
1352 struct obd_uuid obd_uuid;
1357 switch (lcfg->lcfg_command) {
1359 /* modify_mdc_tgts add 0:lustre-clilmv 1:lustre-MDT0000_UUID
1360 * 2:0 3:1 4:lustre-MDT0000-mdc_UUID */
1361 if (LUSTRE_CFG_BUFLEN(lcfg, 1) > sizeof(obd_uuid.uuid)) {
1366 obd_str2uuid(&obd_uuid, lustre_cfg_buf(lcfg, 1));
1368 if (sscanf(lustre_cfg_buf(lcfg, 2), "%d", &index) != 1) {
1372 if (sscanf(lustre_cfg_buf(lcfg, 3), "%d", &gen) != 1) {
1376 rc = lmv_add_target(obd, &obd_uuid, index, gen);
1379 CERROR("Unknown command: %d\n", lcfg->lcfg_command);
1387 static int lmv_statfs(const struct lu_env *env, struct obd_export *exp,
1388 struct obd_statfs *osfs, __u64 max_age, __u32 flags)
1390 struct obd_device *obd = class_exp2obd(exp);
1391 struct lmv_obd *lmv = &obd->u.lmv;
1392 struct obd_statfs *temp;
1396 rc = lmv_check_connect(obd);
1400 temp = kzalloc(sizeof(*temp), GFP_NOFS);
1404 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
1405 if (lmv->tgts[i] == NULL || lmv->tgts[i]->ltd_exp == NULL)
1408 rc = obd_statfs(env, lmv->tgts[i]->ltd_exp, temp,
1411 CERROR("can't stat MDS #%d (%s), error %d\n", i,
1412 lmv->tgts[i]->ltd_exp->exp_obd->obd_name,
1419 /* If the statfs is from mount, it will needs
1420 * retrieve necessary information from MDT0.
1421 * i.e. mount does not need the merged osfs
1423 * And also clients can be mounted as long as
1424 * MDT0 is in service*/
1425 if (flags & OBD_STATFS_FOR_MDT0)
1428 osfs->os_bavail += temp->os_bavail;
1429 osfs->os_blocks += temp->os_blocks;
1430 osfs->os_ffree += temp->os_ffree;
1431 osfs->os_files += temp->os_files;
1440 static int lmv_getstatus(struct obd_export *exp,
1443 struct obd_device *obd = exp->exp_obd;
1444 struct lmv_obd *lmv = &obd->u.lmv;
1447 rc = lmv_check_connect(obd);
1451 rc = md_getstatus(lmv->tgts[0]->ltd_exp, fid);
1455 static int lmv_getxattr(struct obd_export *exp, const struct lu_fid *fid,
1456 u64 valid, const char *name,
1457 const char *input, int input_size, int output_size,
1458 int flags, struct ptlrpc_request **request)
1460 struct obd_device *obd = exp->exp_obd;
1461 struct lmv_obd *lmv = &obd->u.lmv;
1462 struct lmv_tgt_desc *tgt;
1465 rc = lmv_check_connect(obd);
1469 tgt = lmv_find_target(lmv, fid);
1471 return PTR_ERR(tgt);
1473 rc = md_getxattr(tgt->ltd_exp, fid, valid, name, input,
1474 input_size, output_size, flags, request);
1479 static int lmv_setxattr(struct obd_export *exp, const struct lu_fid *fid,
1480 u64 valid, const char *name,
1481 const char *input, int input_size, int output_size,
1482 int flags, __u32 suppgid,
1483 struct ptlrpc_request **request)
1485 struct obd_device *obd = exp->exp_obd;
1486 struct lmv_obd *lmv = &obd->u.lmv;
1487 struct lmv_tgt_desc *tgt;
1490 rc = lmv_check_connect(obd);
1494 tgt = lmv_find_target(lmv, fid);
1496 return PTR_ERR(tgt);
1498 rc = md_setxattr(tgt->ltd_exp, fid, valid, name, input,
1499 input_size, output_size, flags, suppgid,
1505 static int lmv_getattr(struct obd_export *exp, struct md_op_data *op_data,
1506 struct ptlrpc_request **request)
1508 struct obd_device *obd = exp->exp_obd;
1509 struct lmv_obd *lmv = &obd->u.lmv;
1510 struct lmv_tgt_desc *tgt;
1513 rc = lmv_check_connect(obd);
1517 tgt = lmv_find_target(lmv, &op_data->op_fid1);
1519 return PTR_ERR(tgt);
1521 if (op_data->op_flags & MF_GET_MDT_IDX) {
1522 op_data->op_mds = tgt->ltd_idx;
1526 rc = md_getattr(tgt->ltd_exp, op_data, request);
1531 static int lmv_null_inode(struct obd_export *exp, const struct lu_fid *fid)
1533 struct obd_device *obd = exp->exp_obd;
1534 struct lmv_obd *lmv = &obd->u.lmv;
1538 rc = lmv_check_connect(obd);
1542 CDEBUG(D_INODE, "CBDATA for "DFID"\n", PFID(fid));
1545 * With DNE every object can have two locks in different namespaces:
1546 * lookup lock in space of MDT storing direntry and update/open lock in
1547 * space of MDT storing inode.
1549 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
1550 if (lmv->tgts[i] == NULL || lmv->tgts[i]->ltd_exp == NULL)
1552 md_null_inode(lmv->tgts[i]->ltd_exp, fid);
1558 static int lmv_find_cbdata(struct obd_export *exp, const struct lu_fid *fid,
1559 ldlm_iterator_t it, void *data)
1561 struct obd_device *obd = exp->exp_obd;
1562 struct lmv_obd *lmv = &obd->u.lmv;
1566 rc = lmv_check_connect(obd);
1570 CDEBUG(D_INODE, "CBDATA for "DFID"\n", PFID(fid));
1573 * With DNE every object can have two locks in different namespaces:
1574 * lookup lock in space of MDT storing direntry and update/open lock in
1575 * space of MDT storing inode.
1577 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
1578 if (lmv->tgts[i] == NULL || lmv->tgts[i]->ltd_exp == NULL)
1580 rc = md_find_cbdata(lmv->tgts[i]->ltd_exp, fid, it, data);
1588 static int lmv_close(struct obd_export *exp, struct md_op_data *op_data,
1589 struct md_open_data *mod, struct ptlrpc_request **request)
1591 struct obd_device *obd = exp->exp_obd;
1592 struct lmv_obd *lmv = &obd->u.lmv;
1593 struct lmv_tgt_desc *tgt;
1596 rc = lmv_check_connect(obd);
1600 tgt = lmv_find_target(lmv, &op_data->op_fid1);
1602 return PTR_ERR(tgt);
1604 CDEBUG(D_INODE, "CLOSE "DFID"\n", PFID(&op_data->op_fid1));
1605 rc = md_close(tgt->ltd_exp, op_data, mod, request);
1610 *lmv_locate_mds(struct lmv_obd *lmv, struct md_op_data *op_data,
1613 struct lmv_tgt_desc *tgt;
1615 tgt = lmv_find_target(lmv, fid);
1619 op_data->op_mds = tgt->ltd_idx;
1624 static int lmv_create(struct obd_export *exp, struct md_op_data *op_data,
1625 const void *data, int datalen, int mode, __u32 uid,
1626 __u32 gid, cfs_cap_t cap_effective, __u64 rdev,
1627 struct ptlrpc_request **request)
1629 struct obd_device *obd = exp->exp_obd;
1630 struct lmv_obd *lmv = &obd->u.lmv;
1631 struct lmv_tgt_desc *tgt;
1634 rc = lmv_check_connect(obd);
1638 if (!lmv->desc.ld_active_tgt_count)
1641 tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1);
1643 return PTR_ERR(tgt);
1645 rc = lmv_fid_alloc(exp, &op_data->op_fid2, op_data);
1649 CDEBUG(D_INODE, "CREATE '%*s' on "DFID" -> mds #%x\n",
1650 op_data->op_namelen, op_data->op_name, PFID(&op_data->op_fid1),
1653 op_data->op_flags |= MF_MDC_CANCEL_FID1;
1654 rc = md_create(tgt->ltd_exp, op_data, data, datalen, mode, uid, gid,
1655 cap_effective, rdev, request);
1658 if (*request == NULL)
1660 CDEBUG(D_INODE, "Created - "DFID"\n", PFID(&op_data->op_fid2));
1665 static int lmv_done_writing(struct obd_export *exp,
1666 struct md_op_data *op_data,
1667 struct md_open_data *mod)
1669 struct obd_device *obd = exp->exp_obd;
1670 struct lmv_obd *lmv = &obd->u.lmv;
1671 struct lmv_tgt_desc *tgt;
1674 rc = lmv_check_connect(obd);
1678 tgt = lmv_find_target(lmv, &op_data->op_fid1);
1680 return PTR_ERR(tgt);
1682 rc = md_done_writing(tgt->ltd_exp, op_data, mod);
1687 lmv_enqueue_remote(struct obd_export *exp, struct ldlm_enqueue_info *einfo,
1688 struct lookup_intent *it, struct md_op_data *op_data,
1689 struct lustre_handle *lockh, void *lmm, int lmmsize,
1690 __u64 extra_lock_flags)
1692 struct ptlrpc_request *req = it->d.lustre.it_data;
1693 struct obd_device *obd = exp->exp_obd;
1694 struct lmv_obd *lmv = &obd->u.lmv;
1695 struct lustre_handle plock;
1696 struct lmv_tgt_desc *tgt;
1697 struct md_op_data *rdata;
1699 struct mdt_body *body;
1703 body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
1704 LASSERT(body != NULL);
1706 if (!(body->valid & OBD_MD_MDS))
1709 CDEBUG(D_INODE, "REMOTE_ENQUEUE '%s' on "DFID" -> "DFID"\n",
1710 LL_IT2STR(it), PFID(&op_data->op_fid1), PFID(&body->fid1));
1713 * We got LOOKUP lock, but we really need attrs.
1715 pmode = it->d.lustre.it_lock_mode;
1716 LASSERT(pmode != 0);
1717 memcpy(&plock, lockh, sizeof(plock));
1718 it->d.lustre.it_lock_mode = 0;
1719 it->d.lustre.it_data = NULL;
1722 ptlrpc_req_finished(req);
1724 tgt = lmv_find_target(lmv, &fid1);
1730 rdata = kzalloc(sizeof(*rdata), GFP_NOFS);
1736 rdata->op_fid1 = fid1;
1737 rdata->op_bias = MDS_CROSS_REF;
1739 rc = md_enqueue(tgt->ltd_exp, einfo, it, rdata, lockh,
1740 lmm, lmmsize, NULL, extra_lock_flags);
1743 ldlm_lock_decref(&plock, pmode);
1748 lmv_enqueue(struct obd_export *exp, struct ldlm_enqueue_info *einfo,
1749 struct lookup_intent *it, struct md_op_data *op_data,
1750 struct lustre_handle *lockh, void *lmm, int lmmsize,
1751 struct ptlrpc_request **req, __u64 extra_lock_flags)
1753 struct obd_device *obd = exp->exp_obd;
1754 struct lmv_obd *lmv = &obd->u.lmv;
1755 struct lmv_tgt_desc *tgt;
1758 rc = lmv_check_connect(obd);
1762 CDEBUG(D_INODE, "ENQUEUE '%s' on "DFID"\n",
1763 LL_IT2STR(it), PFID(&op_data->op_fid1));
1765 tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1);
1767 return PTR_ERR(tgt);
1769 CDEBUG(D_INODE, "ENQUEUE '%s' on "DFID" -> mds #%d\n",
1770 LL_IT2STR(it), PFID(&op_data->op_fid1), tgt->ltd_idx);
1772 rc = md_enqueue(tgt->ltd_exp, einfo, it, op_data, lockh,
1773 lmm, lmmsize, req, extra_lock_flags);
1775 if (rc == 0 && it && it->it_op == IT_OPEN) {
1776 rc = lmv_enqueue_remote(exp, einfo, it, op_data, lockh,
1777 lmm, lmmsize, extra_lock_flags);
1783 lmv_getattr_name(struct obd_export *exp, struct md_op_data *op_data,
1784 struct ptlrpc_request **request)
1786 struct ptlrpc_request *req = NULL;
1787 struct obd_device *obd = exp->exp_obd;
1788 struct lmv_obd *lmv = &obd->u.lmv;
1789 struct lmv_tgt_desc *tgt;
1790 struct mdt_body *body;
1793 rc = lmv_check_connect(obd);
1797 tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1);
1799 return PTR_ERR(tgt);
1801 CDEBUG(D_INODE, "GETATTR_NAME for %*s on "DFID" -> mds #%d\n",
1802 op_data->op_namelen, op_data->op_name, PFID(&op_data->op_fid1),
1805 rc = md_getattr_name(tgt->ltd_exp, op_data, request);
1809 body = req_capsule_server_get(&(*request)->rq_pill,
1811 LASSERT(body != NULL);
1813 if (body->valid & OBD_MD_MDS) {
1814 struct lu_fid rid = body->fid1;
1816 CDEBUG(D_INODE, "Request attrs for "DFID"\n",
1819 tgt = lmv_find_target(lmv, &rid);
1821 ptlrpc_req_finished(*request);
1822 return PTR_ERR(tgt);
1825 op_data->op_fid1 = rid;
1826 op_data->op_valid |= OBD_MD_FLCROSSREF;
1827 op_data->op_namelen = 0;
1828 op_data->op_name = NULL;
1829 rc = md_getattr_name(tgt->ltd_exp, op_data, &req);
1830 ptlrpc_req_finished(*request);
1837 #define md_op_data_fid(op_data, fl) \
1838 (fl == MF_MDC_CANCEL_FID1 ? &op_data->op_fid1 : \
1839 fl == MF_MDC_CANCEL_FID2 ? &op_data->op_fid2 : \
1840 fl == MF_MDC_CANCEL_FID3 ? &op_data->op_fid3 : \
1841 fl == MF_MDC_CANCEL_FID4 ? &op_data->op_fid4 : \
1844 static int lmv_early_cancel(struct obd_export *exp, struct md_op_data *op_data,
1845 int op_tgt, ldlm_mode_t mode, int bits, int flag)
1847 struct lu_fid *fid = md_op_data_fid(op_data, flag);
1848 struct obd_device *obd = exp->exp_obd;
1849 struct lmv_obd *lmv = &obd->u.lmv;
1850 struct lmv_tgt_desc *tgt;
1851 ldlm_policy_data_t policy = { {0} };
1854 if (!fid_is_sane(fid))
1857 tgt = lmv_find_target(lmv, fid);
1859 return PTR_ERR(tgt);
1861 if (tgt->ltd_idx != op_tgt) {
1862 CDEBUG(D_INODE, "EARLY_CANCEL on "DFID"\n", PFID(fid));
1863 policy.l_inodebits.bits = bits;
1864 rc = md_cancel_unused(tgt->ltd_exp, fid, &policy,
1865 mode, LCF_ASYNC, NULL);
1868 "EARLY_CANCEL skip operation target %d on "DFID"\n",
1870 op_data->op_flags |= flag;
1878 * llite passes fid of an target inode in op_data->op_fid1 and id of directory in
1881 static int lmv_link(struct obd_export *exp, struct md_op_data *op_data,
1882 struct ptlrpc_request **request)
1884 struct obd_device *obd = exp->exp_obd;
1885 struct lmv_obd *lmv = &obd->u.lmv;
1886 struct lmv_tgt_desc *tgt;
1889 rc = lmv_check_connect(obd);
1893 LASSERT(op_data->op_namelen != 0);
1895 CDEBUG(D_INODE, "LINK "DFID":%*s to "DFID"\n",
1896 PFID(&op_data->op_fid2), op_data->op_namelen,
1897 op_data->op_name, PFID(&op_data->op_fid1));
1899 op_data->op_fsuid = from_kuid(&init_user_ns, current_fsuid());
1900 op_data->op_fsgid = from_kgid(&init_user_ns, current_fsgid());
1901 op_data->op_cap = cfs_curproc_cap_pack();
1902 tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid2);
1904 return PTR_ERR(tgt);
1907 * Cancel UPDATE lock on child (fid1).
1909 op_data->op_flags |= MF_MDC_CANCEL_FID2;
1910 rc = lmv_early_cancel(exp, op_data, tgt->ltd_idx, LCK_EX,
1911 MDS_INODELOCK_UPDATE, MF_MDC_CANCEL_FID1);
1915 rc = md_link(tgt->ltd_exp, op_data, request);
1920 static int lmv_rename(struct obd_export *exp, struct md_op_data *op_data,
1921 const char *old, int oldlen, const char *new, int newlen,
1922 struct ptlrpc_request **request)
1924 struct obd_device *obd = exp->exp_obd;
1925 struct lmv_obd *lmv = &obd->u.lmv;
1926 struct lmv_tgt_desc *src_tgt;
1927 struct lmv_tgt_desc *tgt_tgt;
1930 LASSERT(oldlen != 0);
1932 CDEBUG(D_INODE, "RENAME %*s in "DFID" to %*s in "DFID"\n",
1933 oldlen, old, PFID(&op_data->op_fid1),
1934 newlen, new, PFID(&op_data->op_fid2));
1936 rc = lmv_check_connect(obd);
1940 op_data->op_fsuid = from_kuid(&init_user_ns, current_fsuid());
1941 op_data->op_fsgid = from_kgid(&init_user_ns, current_fsgid());
1942 op_data->op_cap = cfs_curproc_cap_pack();
1943 src_tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1);
1944 if (IS_ERR(src_tgt))
1945 return PTR_ERR(src_tgt);
1947 tgt_tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid2);
1948 if (IS_ERR(tgt_tgt))
1949 return PTR_ERR(tgt_tgt);
1951 * LOOKUP lock on src child (fid3) should also be cancelled for
1952 * src_tgt in mdc_rename.
1954 op_data->op_flags |= MF_MDC_CANCEL_FID1 | MF_MDC_CANCEL_FID3;
1957 * Cancel UPDATE locks on tgt parent (fid2), tgt_tgt is its
1960 rc = lmv_early_cancel(exp, op_data, src_tgt->ltd_idx,
1961 LCK_EX, MDS_INODELOCK_UPDATE,
1962 MF_MDC_CANCEL_FID2);
1965 * Cancel LOOKUP locks on tgt child (fid4) for parent tgt_tgt.
1968 rc = lmv_early_cancel(exp, op_data, src_tgt->ltd_idx,
1969 LCK_EX, MDS_INODELOCK_LOOKUP,
1970 MF_MDC_CANCEL_FID4);
1974 * Cancel all the locks on tgt child (fid4).
1977 rc = lmv_early_cancel(exp, op_data, src_tgt->ltd_idx,
1978 LCK_EX, MDS_INODELOCK_FULL,
1979 MF_MDC_CANCEL_FID4);
1982 rc = md_rename(src_tgt->ltd_exp, op_data, old, oldlen,
1983 new, newlen, request);
1987 static int lmv_setattr(struct obd_export *exp, struct md_op_data *op_data,
1988 void *ea, int ealen, void *ea2, int ea2len,
1989 struct ptlrpc_request **request,
1990 struct md_open_data **mod)
1992 struct obd_device *obd = exp->exp_obd;
1993 struct lmv_obd *lmv = &obd->u.lmv;
1994 struct lmv_tgt_desc *tgt;
1997 rc = lmv_check_connect(obd);
2001 CDEBUG(D_INODE, "SETATTR for "DFID", valid 0x%x\n",
2002 PFID(&op_data->op_fid1), op_data->op_attr.ia_valid);
2004 op_data->op_flags |= MF_MDC_CANCEL_FID1;
2005 tgt = lmv_find_target(lmv, &op_data->op_fid1);
2007 return PTR_ERR(tgt);
2009 rc = md_setattr(tgt->ltd_exp, op_data, ea, ealen, ea2,
2010 ea2len, request, mod);
2015 static int lmv_sync(struct obd_export *exp, const struct lu_fid *fid,
2016 struct ptlrpc_request **request)
2018 struct obd_device *obd = exp->exp_obd;
2019 struct lmv_obd *lmv = &obd->u.lmv;
2020 struct lmv_tgt_desc *tgt;
2023 rc = lmv_check_connect(obd);
2027 tgt = lmv_find_target(lmv, fid);
2029 return PTR_ERR(tgt);
2031 rc = md_sync(tgt->ltd_exp, fid, request);
2036 * Adjust a set of pages, each page containing an array of lu_dirpages,
2037 * so that each page can be used as a single logical lu_dirpage.
2039 * A lu_dirpage is laid out as follows, where s = ldp_hash_start,
2040 * e = ldp_hash_end, f = ldp_flags, p = padding, and each "ent" is a
2041 * struct lu_dirent. It has size up to LU_PAGE_SIZE. The ldp_hash_end
2042 * value is used as a cookie to request the next lu_dirpage in a
2043 * directory listing that spans multiple pages (two in this example):
2046 * .|--------v------- -----.
2047 * |s|e|f|p|ent|ent| ... |ent|
2048 * '--|-------------- -----' Each CFS_PAGE contains a single
2049 * '------. lu_dirpage.
2050 * .---------v------- -----.
2051 * |s|e|f|p|ent| 0 | ... | 0 |
2052 * '----------------- -----'
2054 * However, on hosts where the native VM page size (PAGE_CACHE_SIZE) is
2055 * larger than LU_PAGE_SIZE, a single host page may contain multiple
2056 * lu_dirpages. After reading the lu_dirpages from the MDS, the
2057 * ldp_hash_end of the first lu_dirpage refers to the one immediately
2058 * after it in the same CFS_PAGE (arrows simplified for brevity, but
2059 * in general e0==s1, e1==s2, etc.):
2061 * .-------------------- -----.
2062 * |s0|e0|f0|p|ent|ent| ... |ent|
2063 * |---v---------------- -----|
2064 * |s1|e1|f1|p|ent|ent| ... |ent|
2065 * |---v---------------- -----| Here, each CFS_PAGE contains
2066 * ... multiple lu_dirpages.
2067 * |---v---------------- -----|
2068 * |s'|e'|f'|p|ent|ent| ... |ent|
2069 * '---|---------------- -----'
2071 * .----------------------------.
2074 * This structure is transformed into a single logical lu_dirpage as follows:
2076 * - Replace e0 with e' so the request for the next lu_dirpage gets the page
2077 * labeled 'next CFS_PAGE'.
2079 * - Copy the LDF_COLLIDE flag from f' to f0 to correctly reflect whether
2080 * a hash collision with the next page exists.
2082 * - Adjust the lde_reclen of the ending entry of each lu_dirpage to span
2083 * to the first entry of the next lu_dirpage.
2085 #if PAGE_CACHE_SIZE > LU_PAGE_SIZE
2086 static void lmv_adjust_dirpages(struct page **pages, int ncfspgs, int nlupgs)
2090 for (i = 0; i < ncfspgs; i++) {
2091 struct lu_dirpage *dp = kmap(pages[i]);
2092 struct lu_dirpage *first = dp;
2093 struct lu_dirent *end_dirent = NULL;
2094 struct lu_dirent *ent;
2095 __u64 hash_end = dp->ldp_hash_end;
2096 __u32 flags = dp->ldp_flags;
2098 while (--nlupgs > 0) {
2099 ent = lu_dirent_start(dp);
2100 for (end_dirent = ent; ent != NULL;
2101 end_dirent = ent, ent = lu_dirent_next(ent))
2104 /* Advance dp to next lu_dirpage. */
2105 dp = (struct lu_dirpage *)((char *)dp + LU_PAGE_SIZE);
2107 /* Check if we've reached the end of the CFS_PAGE. */
2108 if (!((unsigned long)dp & ~CFS_PAGE_MASK))
2111 /* Save the hash and flags of this lu_dirpage. */
2112 hash_end = dp->ldp_hash_end;
2113 flags = dp->ldp_flags;
2115 /* Check if lu_dirpage contains no entries. */
2119 /* Enlarge the end entry lde_reclen from 0 to
2120 * first entry of next lu_dirpage. */
2121 LASSERT(le16_to_cpu(end_dirent->lde_reclen) == 0);
2122 end_dirent->lde_reclen =
2123 cpu_to_le16((char *)(dp->ldp_entries) -
2124 (char *)end_dirent);
2127 first->ldp_hash_end = hash_end;
2128 first->ldp_flags &= ~cpu_to_le32(LDF_COLLIDE);
2129 first->ldp_flags |= flags & cpu_to_le32(LDF_COLLIDE);
2133 LASSERTF(nlupgs == 0, "left = %d", nlupgs);
2136 #define lmv_adjust_dirpages(pages, ncfspgs, nlupgs) do {} while (0)
2137 #endif /* PAGE_CACHE_SIZE > LU_PAGE_SIZE */
2139 static int lmv_readpage(struct obd_export *exp, struct md_op_data *op_data,
2140 struct page **pages, struct ptlrpc_request **request)
2142 struct obd_device *obd = exp->exp_obd;
2143 struct lmv_obd *lmv = &obd->u.lmv;
2144 __u64 offset = op_data->op_offset;
2146 int ncfspgs; /* pages read in PAGE_CACHE_SIZE */
2147 int nlupgs; /* pages read in LU_PAGE_SIZE */
2148 struct lmv_tgt_desc *tgt;
2150 rc = lmv_check_connect(obd);
2154 CDEBUG(D_INODE, "READPAGE at %#llx from "DFID"\n",
2155 offset, PFID(&op_data->op_fid1));
2157 tgt = lmv_find_target(lmv, &op_data->op_fid1);
2159 return PTR_ERR(tgt);
2161 rc = md_readpage(tgt->ltd_exp, op_data, pages, request);
2165 ncfspgs = ((*request)->rq_bulk->bd_nob_transferred + PAGE_CACHE_SIZE - 1)
2166 >> PAGE_CACHE_SHIFT;
2167 nlupgs = (*request)->rq_bulk->bd_nob_transferred >> LU_PAGE_SHIFT;
2168 LASSERT(!((*request)->rq_bulk->bd_nob_transferred & ~LU_PAGE_MASK));
2169 LASSERT(ncfspgs > 0 && ncfspgs <= op_data->op_npages);
2171 CDEBUG(D_INODE, "read %d(%d)/%d pages\n", ncfspgs, nlupgs,
2172 op_data->op_npages);
2174 lmv_adjust_dirpages(pages, ncfspgs, nlupgs);
2179 static int lmv_unlink(struct obd_export *exp, struct md_op_data *op_data,
2180 struct ptlrpc_request **request)
2182 struct obd_device *obd = exp->exp_obd;
2183 struct lmv_obd *lmv = &obd->u.lmv;
2184 struct lmv_tgt_desc *tgt = NULL;
2185 struct mdt_body *body;
2188 rc = lmv_check_connect(obd);
2192 /* Send unlink requests to the MDT where the child is located */
2193 if (likely(!fid_is_zero(&op_data->op_fid2)))
2194 tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid2);
2196 tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1);
2198 return PTR_ERR(tgt);
2200 op_data->op_fsuid = from_kuid(&init_user_ns, current_fsuid());
2201 op_data->op_fsgid = from_kgid(&init_user_ns, current_fsgid());
2202 op_data->op_cap = cfs_curproc_cap_pack();
2205 * If child's fid is given, cancel unused locks for it if it is from
2206 * another export than parent.
2208 * LOOKUP lock for child (fid3) should also be cancelled on parent
2209 * tgt_tgt in mdc_unlink().
2211 op_data->op_flags |= MF_MDC_CANCEL_FID1 | MF_MDC_CANCEL_FID3;
2214 * Cancel FULL locks on child (fid3).
2216 rc = lmv_early_cancel(exp, op_data, tgt->ltd_idx, LCK_EX,
2217 MDS_INODELOCK_FULL, MF_MDC_CANCEL_FID3);
2222 CDEBUG(D_INODE, "unlink with fid="DFID"/"DFID" -> mds #%d\n",
2223 PFID(&op_data->op_fid1), PFID(&op_data->op_fid2), tgt->ltd_idx);
2225 rc = md_unlink(tgt->ltd_exp, op_data, request);
2226 if (rc != 0 && rc != -EREMOTE)
2229 body = req_capsule_server_get(&(*request)->rq_pill, &RMF_MDT_BODY);
2233 /* Not cross-ref case, just get out of here. */
2234 if (likely(!(body->valid & OBD_MD_MDS)))
2237 CDEBUG(D_INODE, "%s: try unlink to another MDT for "DFID"\n",
2238 exp->exp_obd->obd_name, PFID(&body->fid1));
2240 /* This is a remote object, try remote MDT, Note: it may
2241 * try more than 1 time here, Considering following case
2242 * /mnt/lustre is root on MDT0, remote1 is on MDT1
2243 * 1. Initially A does not know where remote1 is, it send
2244 * unlink RPC to MDT0, MDT0 return -EREMOTE, it will
2245 * resend unlink RPC to MDT1 (retry 1st time).
2247 * 2. During the unlink RPC in flight,
2248 * client B mv /mnt/lustre/remote1 /mnt/lustre/remote2
2249 * and create new remote1, but on MDT0
2251 * 3. MDT1 get unlink RPC(from A), then do remote lock on
2252 * /mnt/lustre, then lookup get fid of remote1, and find
2253 * it is remote dir again, and replay -EREMOTE again.
2255 * 4. Then A will resend unlink RPC to MDT0. (retry 2nd times).
2257 * In theory, it might try unlimited time here, but it should
2258 * be very rare case. */
2259 op_data->op_fid2 = body->fid1;
2260 ptlrpc_req_finished(*request);
2266 static int lmv_precleanup(struct obd_device *obd, enum obd_cleanup_stage stage)
2268 struct lmv_obd *lmv = &obd->u.lmv;
2271 case OBD_CLEANUP_EARLY:
2272 /* XXX: here should be calling obd_precleanup() down to
2275 case OBD_CLEANUP_EXPORTS:
2276 fld_client_debugfs_fini(&lmv->lmv_fld);
2277 lprocfs_obd_cleanup(obd);
2285 static int lmv_get_info(const struct lu_env *env, struct obd_export *exp,
2286 __u32 keylen, void *key, __u32 *vallen, void *val,
2287 struct lov_stripe_md *lsm)
2289 struct obd_device *obd;
2290 struct lmv_obd *lmv;
2293 obd = class_exp2obd(exp);
2295 CDEBUG(D_IOCTL, "Invalid client cookie %#llx\n",
2296 exp->exp_handle.h_cookie);
2301 if (keylen >= strlen("remote_flag") && !strcmp(key, "remote_flag")) {
2302 struct lmv_tgt_desc *tgt;
2305 rc = lmv_check_connect(obd);
2309 LASSERT(*vallen == sizeof(__u32));
2310 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2313 * All tgts should be connected when this gets called.
2315 if (tgt == NULL || tgt->ltd_exp == NULL)
2318 if (!obd_get_info(env, tgt->ltd_exp, keylen, key,
2323 } else if (KEY_IS(KEY_MAX_EASIZE) ||
2324 KEY_IS(KEY_DEFAULT_EASIZE) ||
2325 KEY_IS(KEY_CONN_DATA)) {
2326 rc = lmv_check_connect(obd);
2331 * Forwarding this request to first MDS, it should know LOV
2334 rc = obd_get_info(env, lmv->tgts[0]->ltd_exp, keylen, key,
2336 if (!rc && KEY_IS(KEY_CONN_DATA))
2337 exp->exp_connect_data = *(struct obd_connect_data *)val;
2339 } else if (KEY_IS(KEY_TGT_COUNT)) {
2340 *((int *)val) = lmv->desc.ld_tgt_count;
2344 CDEBUG(D_IOCTL, "Invalid key\n");
2348 static int lmv_set_info_async(const struct lu_env *env, struct obd_export *exp,
2349 u32 keylen, void *key, u32 vallen,
2350 void *val, struct ptlrpc_request_set *set)
2352 struct lmv_tgt_desc *tgt;
2353 struct obd_device *obd;
2354 struct lmv_obd *lmv;
2357 obd = class_exp2obd(exp);
2359 CDEBUG(D_IOCTL, "Invalid client cookie %#llx\n",
2360 exp->exp_handle.h_cookie);
2365 if (KEY_IS(KEY_READ_ONLY) || KEY_IS(KEY_FLUSH_CTX)) {
2368 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2371 if (tgt == NULL || tgt->ltd_exp == NULL)
2374 err = obd_set_info_async(env, tgt->ltd_exp,
2375 keylen, key, vallen, val, set);
2386 static int lmv_packmd(struct obd_export *exp, struct lov_mds_md **lmmp,
2387 struct lov_stripe_md *lsm)
2389 struct obd_device *obd = class_exp2obd(exp);
2390 struct lmv_obd *lmv = &obd->u.lmv;
2391 struct lmv_stripe_md *meap;
2392 struct lmv_stripe_md *lsmp;
2396 mea_size = lmv_get_easize(lmv);
2400 if (*lmmp && !lsm) {
2406 if (*lmmp == NULL) {
2407 *lmmp = libcfs_kvzalloc(mea_size, GFP_NOFS);
2415 lsmp = (struct lmv_stripe_md *)lsm;
2416 meap = (struct lmv_stripe_md *)*lmmp;
2418 if (lsmp->mea_magic != MEA_MAGIC_LAST_CHAR &&
2419 lsmp->mea_magic != MEA_MAGIC_ALL_CHARS)
2422 meap->mea_magic = cpu_to_le32(lsmp->mea_magic);
2423 meap->mea_count = cpu_to_le32(lsmp->mea_count);
2424 meap->mea_master = cpu_to_le32(lsmp->mea_master);
2426 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2427 meap->mea_ids[i] = lsmp->mea_ids[i];
2428 fid_cpu_to_le(&meap->mea_ids[i], &lsmp->mea_ids[i]);
2434 static int lmv_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
2435 struct lov_mds_md *lmm, int lmm_size)
2437 struct obd_device *obd = class_exp2obd(exp);
2438 struct lmv_stripe_md **tmea = (struct lmv_stripe_md **)lsmp;
2439 struct lmv_stripe_md *mea = (struct lmv_stripe_md *)lmm;
2440 struct lmv_obd *lmv = &obd->u.lmv;
2445 mea_size = lmv_get_easize(lmv);
2449 if (*lsmp != NULL && lmm == NULL) {
2455 LASSERT(mea_size == lmm_size);
2457 *tmea = libcfs_kvzalloc(mea_size, GFP_NOFS);
2464 if (mea->mea_magic == MEA_MAGIC_LAST_CHAR ||
2465 mea->mea_magic == MEA_MAGIC_ALL_CHARS ||
2466 mea->mea_magic == MEA_MAGIC_HASH_SEGMENT) {
2467 magic = le32_to_cpu(mea->mea_magic);
2470 * Old mea is not handled here.
2472 CERROR("Old not supportable EA is found\n");
2476 (*tmea)->mea_magic = magic;
2477 (*tmea)->mea_count = le32_to_cpu(mea->mea_count);
2478 (*tmea)->mea_master = le32_to_cpu(mea->mea_master);
2480 for (i = 0; i < (*tmea)->mea_count; i++) {
2481 (*tmea)->mea_ids[i] = mea->mea_ids[i];
2482 fid_le_to_cpu(&(*tmea)->mea_ids[i], &(*tmea)->mea_ids[i]);
2487 static int lmv_cancel_unused(struct obd_export *exp, const struct lu_fid *fid,
2488 ldlm_policy_data_t *policy, ldlm_mode_t mode,
2489 ldlm_cancel_flags_t flags, void *opaque)
2491 struct obd_device *obd = exp->exp_obd;
2492 struct lmv_obd *lmv = &obd->u.lmv;
2497 LASSERT(fid != NULL);
2499 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2500 if (lmv->tgts[i] == NULL || lmv->tgts[i]->ltd_exp == NULL ||
2501 lmv->tgts[i]->ltd_active == 0)
2504 err = md_cancel_unused(lmv->tgts[i]->ltd_exp, fid,
2505 policy, mode, flags, opaque);
2512 static int lmv_set_lock_data(struct obd_export *exp, __u64 *lockh, void *data,
2515 struct lmv_obd *lmv = &exp->exp_obd->u.lmv;
2518 rc = md_set_lock_data(lmv->tgts[0]->ltd_exp, lockh, data, bits);
2522 static ldlm_mode_t lmv_lock_match(struct obd_export *exp, __u64 flags,
2523 const struct lu_fid *fid, ldlm_type_t type,
2524 ldlm_policy_data_t *policy, ldlm_mode_t mode,
2525 struct lustre_handle *lockh)
2527 struct obd_device *obd = exp->exp_obd;
2528 struct lmv_obd *lmv = &obd->u.lmv;
2532 CDEBUG(D_INODE, "Lock match for "DFID"\n", PFID(fid));
2535 * With CMD every object can have two locks in different namespaces:
2536 * lookup lock in space of mds storing direntry and update/open lock in
2537 * space of mds storing inode. Thus we check all targets, not only that
2538 * one fid was created in.
2540 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2541 if (lmv->tgts[i] == NULL ||
2542 lmv->tgts[i]->ltd_exp == NULL ||
2543 lmv->tgts[i]->ltd_active == 0)
2546 rc = md_lock_match(lmv->tgts[i]->ltd_exp, flags, fid,
2547 type, policy, mode, lockh);
2555 static int lmv_get_lustre_md(struct obd_export *exp,
2556 struct ptlrpc_request *req,
2557 struct obd_export *dt_exp,
2558 struct obd_export *md_exp,
2559 struct lustre_md *md)
2561 struct lmv_obd *lmv = &exp->exp_obd->u.lmv;
2563 return md_get_lustre_md(lmv->tgts[0]->ltd_exp, req, dt_exp, md_exp, md);
2566 static int lmv_free_lustre_md(struct obd_export *exp, struct lustre_md *md)
2568 struct obd_device *obd = exp->exp_obd;
2569 struct lmv_obd *lmv = &obd->u.lmv;
2572 obd_free_memmd(exp, (void *)&md->mea);
2573 return md_free_lustre_md(lmv->tgts[0]->ltd_exp, md);
2576 static int lmv_set_open_replay_data(struct obd_export *exp,
2577 struct obd_client_handle *och,
2578 struct lookup_intent *it)
2580 struct obd_device *obd = exp->exp_obd;
2581 struct lmv_obd *lmv = &obd->u.lmv;
2582 struct lmv_tgt_desc *tgt;
2584 tgt = lmv_find_target(lmv, &och->och_fid);
2586 return PTR_ERR(tgt);
2588 return md_set_open_replay_data(tgt->ltd_exp, och, it);
2591 static int lmv_clear_open_replay_data(struct obd_export *exp,
2592 struct obd_client_handle *och)
2594 struct obd_device *obd = exp->exp_obd;
2595 struct lmv_obd *lmv = &obd->u.lmv;
2596 struct lmv_tgt_desc *tgt;
2598 tgt = lmv_find_target(lmv, &och->och_fid);
2600 return PTR_ERR(tgt);
2602 return md_clear_open_replay_data(tgt->ltd_exp, och);
2605 static int lmv_get_remote_perm(struct obd_export *exp,
2606 const struct lu_fid *fid,
2607 __u32 suppgid, struct ptlrpc_request **request)
2609 struct obd_device *obd = exp->exp_obd;
2610 struct lmv_obd *lmv = &obd->u.lmv;
2611 struct lmv_tgt_desc *tgt;
2614 rc = lmv_check_connect(obd);
2618 tgt = lmv_find_target(lmv, fid);
2620 return PTR_ERR(tgt);
2622 rc = md_get_remote_perm(tgt->ltd_exp, fid, suppgid, request);
2626 static int lmv_intent_getattr_async(struct obd_export *exp,
2627 struct md_enqueue_info *minfo,
2628 struct ldlm_enqueue_info *einfo)
2630 struct md_op_data *op_data = &minfo->mi_data;
2631 struct obd_device *obd = exp->exp_obd;
2632 struct lmv_obd *lmv = &obd->u.lmv;
2633 struct lmv_tgt_desc *tgt = NULL;
2636 rc = lmv_check_connect(obd);
2640 tgt = lmv_find_target(lmv, &op_data->op_fid1);
2642 return PTR_ERR(tgt);
2644 rc = md_intent_getattr_async(tgt->ltd_exp, minfo, einfo);
2648 static int lmv_revalidate_lock(struct obd_export *exp, struct lookup_intent *it,
2649 struct lu_fid *fid, __u64 *bits)
2651 struct obd_device *obd = exp->exp_obd;
2652 struct lmv_obd *lmv = &obd->u.lmv;
2653 struct lmv_tgt_desc *tgt;
2656 rc = lmv_check_connect(obd);
2660 tgt = lmv_find_target(lmv, fid);
2662 return PTR_ERR(tgt);
2664 rc = md_revalidate_lock(tgt->ltd_exp, it, fid, bits);
2669 * For lmv, only need to send request to master MDT, and the master MDT will
2670 * process with other slave MDTs. The only exception is Q_GETOQUOTA for which
2671 * we directly fetch data from the slave MDTs.
2673 static int lmv_quotactl(struct obd_device *unused, struct obd_export *exp,
2674 struct obd_quotactl *oqctl)
2676 struct obd_device *obd = class_exp2obd(exp);
2677 struct lmv_obd *lmv = &obd->u.lmv;
2678 struct lmv_tgt_desc *tgt = lmv->tgts[0];
2680 __u64 curspace, curinodes;
2682 if (!lmv->desc.ld_tgt_count || !tgt->ltd_active) {
2683 CERROR("master lmv inactive\n");
2687 if (oqctl->qc_cmd != Q_GETOQUOTA) {
2688 rc = obd_quotactl(tgt->ltd_exp, oqctl);
2692 curspace = curinodes = 0;
2693 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2698 if (tgt == NULL || tgt->ltd_exp == NULL || tgt->ltd_active == 0)
2700 if (!tgt->ltd_active) {
2701 CDEBUG(D_HA, "mdt %d is inactive.\n", i);
2705 err = obd_quotactl(tgt->ltd_exp, oqctl);
2707 CERROR("getquota on mdt %d failed. %d\n", i, err);
2711 curspace += oqctl->qc_dqblk.dqb_curspace;
2712 curinodes += oqctl->qc_dqblk.dqb_curinodes;
2715 oqctl->qc_dqblk.dqb_curspace = curspace;
2716 oqctl->qc_dqblk.dqb_curinodes = curinodes;
2721 static int lmv_quotacheck(struct obd_device *unused, struct obd_export *exp,
2722 struct obd_quotactl *oqctl)
2724 struct obd_device *obd = class_exp2obd(exp);
2725 struct lmv_obd *lmv = &obd->u.lmv;
2726 struct lmv_tgt_desc *tgt;
2729 for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
2733 if (tgt == NULL || tgt->ltd_exp == NULL || !tgt->ltd_active) {
2734 CERROR("lmv idx %d inactive\n", i);
2738 err = obd_quotacheck(tgt->ltd_exp, oqctl);
2746 static struct obd_ops lmv_obd_ops = {
2747 .o_owner = THIS_MODULE,
2748 .o_setup = lmv_setup,
2749 .o_cleanup = lmv_cleanup,
2750 .o_precleanup = lmv_precleanup,
2751 .o_process_config = lmv_process_config,
2752 .o_connect = lmv_connect,
2753 .o_disconnect = lmv_disconnect,
2754 .o_statfs = lmv_statfs,
2755 .o_get_info = lmv_get_info,
2756 .o_set_info_async = lmv_set_info_async,
2757 .o_packmd = lmv_packmd,
2758 .o_unpackmd = lmv_unpackmd,
2759 .o_notify = lmv_notify,
2760 .o_get_uuid = lmv_get_uuid,
2761 .o_iocontrol = lmv_iocontrol,
2762 .o_quotacheck = lmv_quotacheck,
2763 .o_quotactl = lmv_quotactl
2766 static struct md_ops lmv_md_ops = {
2767 .m_getstatus = lmv_getstatus,
2768 .m_null_inode = lmv_null_inode,
2769 .m_find_cbdata = lmv_find_cbdata,
2770 .m_close = lmv_close,
2771 .m_create = lmv_create,
2772 .m_done_writing = lmv_done_writing,
2773 .m_enqueue = lmv_enqueue,
2774 .m_getattr = lmv_getattr,
2775 .m_getxattr = lmv_getxattr,
2776 .m_getattr_name = lmv_getattr_name,
2777 .m_intent_lock = lmv_intent_lock,
2779 .m_rename = lmv_rename,
2780 .m_setattr = lmv_setattr,
2781 .m_setxattr = lmv_setxattr,
2783 .m_readpage = lmv_readpage,
2784 .m_unlink = lmv_unlink,
2785 .m_init_ea_size = lmv_init_ea_size,
2786 .m_cancel_unused = lmv_cancel_unused,
2787 .m_set_lock_data = lmv_set_lock_data,
2788 .m_lock_match = lmv_lock_match,
2789 .m_get_lustre_md = lmv_get_lustre_md,
2790 .m_free_lustre_md = lmv_free_lustre_md,
2791 .m_set_open_replay_data = lmv_set_open_replay_data,
2792 .m_clear_open_replay_data = lmv_clear_open_replay_data,
2793 .m_get_remote_perm = lmv_get_remote_perm,
2794 .m_intent_getattr_async = lmv_intent_getattr_async,
2795 .m_revalidate_lock = lmv_revalidate_lock
2798 static int __init lmv_init(void)
2800 struct lprocfs_static_vars lvars;
2803 lprocfs_lmv_init_vars(&lvars);
2805 rc = class_register_type(&lmv_obd_ops, &lmv_md_ops,
2806 LUSTRE_LMV_NAME, NULL);
2810 static void lmv_exit(void)
2812 class_unregister_type(LUSTRE_LMV_NAME);
2815 MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
2816 MODULE_DESCRIPTION("Lustre Logical Metadata Volume OBD driver");
2817 MODULE_LICENSE("GPL");
2819 module_init(lmv_init);
2820 module_exit(lmv_exit);